...
As reported, we believe that the fluctuation of datasets is due the circulation problem. Assuming datasets from previous experiment experiencing similar problem, we would consider to proceed with data analysis and comparison.
Figure 1. 10 datasets of experiment (x) 
with the ensemble average (red line) for the old setup.
Figure 2. 10 datasets of experiment (x) with the ensemble average (red line) for the new setup.
Figure 3 illustrates the effect of different setups of tube flocculator. Using similar initial parameters, the new setup shows a faster drop than the previous setup. This could be an indicator that the new setup has more bigger flocs during the settling state than the previous one. Data smoothing is applied to eliminate fluctuation due to recirculation problem that previously indicated. Only one dataset is chosen to illustrate the point.
Figure 3. Ensemble average datasets for both old and new setups.